Inspiration
Job interviews can be stressful for anyone, but they can be especially challenging for international students, newcomers, and non-native English speakers. Many candidates worry that their accent or communication style may affect how they are perceived, even when they have the right skills and experience. We wanted to build a tool that helps people practice interviews with confidence while promoting fairer communication and equal opportunities.
What it does
EquaVoice AI is an AI-powered interview coach that allows users to practice mock interviews through voice conversations. It converts spoken responses into text, analyzes communication skills such as clarity, confidence, professionalism, speaking pace, and filler words, then provides personalized feedback to help users improve. Interview sessions are saved so users can review their progress over time.
How we built it
We built the frontend with React and TypeScript, and the backend with Flask and Python. Whisper is used for speech-to-text conversion, OpenAI models analyze interview responses and generate personalized feedback, and MongoDB Atlas stores interview sessions, transcripts, scores, and progress data. Our architecture connects voice input, AI analysis, and data storage into one seamless interview practice experience.
Challenges we ran into
One of our biggest challenges was integrating multiple AI services into a reliable end-to-end workflow. We needed to make sure the frontend, backend, AI services, and database could communicate using a consistent data structure. Another challenge was designing feedback that focuses on improving communication effectiveness rather than encouraging users to change their identity or simply "sound more native."
Accomplishments that we're proud of
We are proud of building a complete AI interview coaching experience within a short hackathon timeframe, from voice recording and transcription to AI-generated feedback and progress tracking. Beyond the technical implementation, we created a project that uses AI to support more inclusive and accessible interview preparation for people from diverse backgrounds.
What we learned
Through this project, we learned how to combine speech recognition, large language models, frontend development, backend APIs, and cloud databases into a complete application. We also learned that responsible AI is not only about building accurate models, but also about creating systems that are transparent, inclusive, and focused on helping real users.
What's next for EquaVoice AI - Different Voices. Equal Opportunities.
We want to continue improving EquaVoice AI with features such as industry-specific interview simulations, multilingual support, real-time coaching, personalized learning plans, and deeper progress analytics. Our vision is to make high-quality interview preparation accessible to everyone and help candidates be evaluated based on their skills, ideas, and potential, rather than their accent or background.
Built With
- google-gemini
- mongodb
- mongoose
- next.js
- react
- shadcn/ui
- tailwindcss
- typescript
- vercel
- web-speech-api
Log in or sign up for Devpost to join the conversation.